Human Genome to Pipefitting?

Robert: Well, welcome everybody to the Contech Exec Podcast. Today I've got a special guest, David Pesiji Basigi, and if I get it right,

David Basiji: Ha ha ha.

Robert: from Pipe Server. Now, I'm gonna tell you folks, this is a a little bit of a little bit of deviation. Not not really, I mean it's still construction technology, but David has got a unique path to the world of construction technology, and the minute that I heard about it, I really wanted to get David in here and get to talking to him about it because what he's got going on is really for builders. it really is. And it's the first technology I think that we're gonna d talk about that, you know, actually gets into this whole, you know, the fabrication portion of the world. But David, welcome to the welcome to the Contact Exec podcast. I'm glad you're here.

David Basiji: Thanks, Robert. It's great to be here.

Robert: Well, so obviously you've got a unique background and I can't tell the story as well as probably you can. So I want to kind of give us a little history and don't leave out, you know, any of the the the the fun details about how you came from the world of, you know, chemistry, biology, whatever that world was, and got yourself into construction technology. So I'd wanna hear the story.

David Basiji: Sure, yeah, actually it's it it's relevant because that's when I met Kelly Dillon, who was the founder of Pipe Server, way back in you know the early actually the late nineties. So I was studying bioengineering and ended up you know getting a getting a degree there, working in a in a lab on on at the time was the Human Genome Project. So sequencing all the DNA in in the in the human body. And and I was specific I'm a tool maker. I was specifically on the instrumentation side. And when they started that project, they had to take the existing DNA sequencing technology and make it about ten thousand times faster in in over the course of the twenty year project to to get it done. So the f really the first like sixteen years of that whole project was just making instruments, making stuff faster. And that was what I was working on. And

Robert: Let's say now wait a minute. You say making instruments, you're talking about the the actual tools that you know, go into the cells and and look at things and do kind of explain a little bit that instrumentation kind of thing.

David Basiji: So that the so basically it was DNA sequencers. So so these are machines that crack open the cells, take the DNA out, ampli what's called amplify it, which is make make millions and millions of copies on it of it because they they can't detect or at the time they couldn't detect a single molecule. They weren't sensitive enough. So they'd make millions of copies of of it. and then they would try and figure out the sequence of the four DNA base pairs, A G C T. And and and so I was working on on machines to do that kind of thing. And Kelly came in well you know, into that lab one summer. He at the time he was working working for Microsoft and and he was between projects and he read this article in the in the Wall Street Journal about the human genome project and and he snuck into our secure building. And threw the Wall Street Journal on my professor's desk and said, I got the summer off, I'm a software engineer, I'll work for free. just put me to work. This is such a cool project. I I want to be involved. And and then my my professor, you know, being being who he was, he's like, Free labor? Absolutely, no problem. So he

Robert: Right.

David Basiji: he walks Kelly around and and Kelly and I met that way and and I happen to be deep in the lab with all the lights out and the the lasers on and the mirror spinning and and and he walked in and and I had my goggles on and and you know my professor says, Kelly this is Dave, Dave this is Kelly and and and we hit it off right away. and so we we sort of formed a friendship way back when and then you know, long story short, I graduated, started a company, and and that company was it wasn't building DNA sequencers, it was building high speed microscopes, automated microscopes. And those

Robert: Okay. Which is yeah, yeah, which is kind of all in that same vein. You needed that yeah that kind of tech, right?

David Basiji: Exactly, exactly. And and that was that was a cool tack that, you know, would image cells right in the bodily fluids. You didn't have to put on a slide. You could just take some blood or or mucus or whatever other bodily fluid you were interested in and there's cells in it and you could put it into that machine and it would it would shoot through and take pictures on the fly. And so we built that company up and that started in ninety nine and I sold it in twenty eleven. started out as chief technology officer and then took over as CEO about halfway halfway through and built it up, got it profitable, put out multiple products, you know, ultimately sold it. And and it was a great experience. You know, it was not just a technology experience. It was learning business and deal making and sales and marketing and and all that stuff. And it was crazy. It was a it was a wild ride. And and Kelly in right about the time I sold my company, started PipeServer. And and so, you know, about every quarter, every six months or so, he'd call me up after a few years when once things got rolling. And, you know, he he'd have some challenge or some opportunity. And because I'd already been through it, you know, we'd spitball and and and and he'd say, what do you think about that? What you think about this? And so I had a lot of familiarity with PipeServer over the years as he as he grew it. and but pipe server started and for most of its life was you know just software for one brand of pipe cutter. They this is a company that was started by you know here in Washington State. It was started by some ex Vernon Tool folks. after Vernon got bought by Lincoln Electric, they left and and they wanted to build sort of a next generation pipe cutter that was all competitive. Computerized and and they needed somebody to write software for it. And they found Kelly, and so he did that. And in the process, he realized look, this is a lot more than just you know, like programming a CNC mill. You know, this is the there's a bigger problem here. And that problem is how do you get data from at the time mostly CAD programs into these machines without somebody having to, you know, write up a spool sheet. and then take that spool sheet over to the machine and then

Robert: Mm.

David Basiji: you know punch in all the dimensions and the locations of the holes and the sizes and the bevels and all that stuff and hopefully not make a mistake and and you know one part at a time. so the machines are really fast, right? They're computerized C and C plasma torches. You know, they once they get cut in, you know, that part's cut in a in a minute. But but all the upstream stuff of getting the data from CAD and and over into the machine and and all that was was a nightmare.

Robert: I have the same problem.

David Basiji: So all the bottlenecks.

Robert: I have the same problem. So I I have a small company, a little MTCO carpentry. We build I build custom furniture and I build custom acoustic products. You can see some of the stuff behind me. Right.

David Basiji: Uh-huh.

Robert: I have the same problem. I have a design, right? Where it's it's on a it's on a well, I do it in sketch up or whatever, but it's a it's a full design and sketch up. But I have to create these cut lists in these things, right, for each individual piece instead of it being able to pull it right off of the drawing.

David Basiji: Yeah. And so you know the pain. And

Robert: Yeah.

David Basiji: and in, you know, in commercial construction, there's there's not a lot of standardization. Maybe you can standardize a skid or or a a carrier or something, but but you know, I mean every building's different. That's why we have architects, right? And and and so all the jobs are different and every cut list is different. And it's not like you've got a you know, a small set of products and you you feel the pain once of making those cut lists, but after that it's just turned the crank. You know, I'm I'm gonna make a hundred copies of this and sell to a hundred different customers. It's not like that in construction. Every building is different, every floor is different, every job's different. So this problem just is very, very painful and it never goes away.

Robert: Right. It's it's a massive yeah, it's a massive point 'cause not only do you have all the H V A C that you have to do, then you've also got your plumbing pipe, you've got your electrical pipe, you've got those conduits. Like anything that has got some sort of pipe or conduit or something, needs to have something like this because if it needs to be prefabbed or cut, like you don't want to be going through this, you know, these iterations of the way we've been doing it for the last fifty years.

David Basiji: Exactly. Exactly. And you know, and people have gone from paper to PDF and people have gone, you know, from from handwritten cut list to spreadsheets and and some of these tools can, you know, if they're computerized, they can pull pull the Excel files in and that sort of thing. But, you know, you still have this fundamental problem that it's a static thing. You can you can accidentally cut the same list twice. you can accidentally cut an old list, you know, when some of the stuff has changed and you

Robert: Yeah. yeah.

David Basiji: You know, you weren't aware of it, didn't come from design fast enough.

Robert: Yeah, the the change order didn't quite come through just yet and it's not showing in the in the new set of drawing or the new new cut list.

David Basiji: Exactly. And and and even once it's cut, how do you know it's cut? You know, how does how do other people know where the job is, you know, in its process? So

Robert: Mm-hmm.

David Basiji: going from paper to digital is good, but you still have a lot of problems that are inherent in the paper process. So you know, Kelly came to me in twenty nineteen and said, Hey, you know, I'd like to I'd like to expand pipe server, you know, beyond just this one brand of pipe cutter and onto all the pipe cutters. and on top of that, at the time, you know, BIM was was getting a lot of traction, Revit was getting a lot of traction. And so we needed to sort of meet that new world as well. And it's not just the old school CAD programs, AutoCAD and SOLIDWORKS and and so on. So so I joined, you know, I I I'd been doing a ton of travel for for work. I'd I'd sold my company in in twenty eleven, worked for the choir for for a few years, then worked with one of my old investors in another biotech for a few years and I was just traveling you know crazy amounts and and I I needed to get off that treadmill. and this sounded really interesting in in part because it was a shift. You know it was a I I like learning new stuff and this is a different industry and a lot to learn. And on top of that you know it was a it's it it it looked a lot like the environment I was in with my company where there were one or two big players out there and and a bunch of little ones. And and, you know, with the right technology and the right strategy, you could really carve out a niche for yourself and and compete against some big players successfully and and and you know really grow the business in in in this kind of context. Something I'd done before. and and so it looked like you know, similar challenge here. So I came in and and we actually expanded beyond, you know, just getting onto pipe cutters. We we realized we gotta we gotta get into every machine in the fab shop. so you know, not just cutting pipe, but cutting anything you'd cut with a saw or a snapper, you know, not only computerized tools but manual tools, bending, grooving, all

Robert: Bendings. Yeah. Shearing. Yeah, all of that, yeah.

David Basiji: all that stuff. And and welding even. And and so, you know, every every every tool you might see in a fab shop, we wanted to have, you know, some some finger in that pie. because there's a there's a network effect, right? If you start getting all the tools in the same system and and all talking to each other and and and talking to the to the BIM up upstream and talking to the field downstream, then things really get efficient really fast.

Robert: Right. It's a ma it's a manufacturing operation, right? You start at one end with the raw materials and you send it kind of down the line, basically, and making sure that it's you got it this you got a cut here, and then you've got a bend over here, and then you've got holes that go in over here, and then you've got some welds that go in over here. Well, the sequence of those is very, very important.

David Basiji: Yep, yep. And there's a lot of you know the the the BIM model is is supposed to be as built, right? That that needs to be what is handed off to the owner and you know it's it's the holy grail of what if you punch a hole hole in this wall, you're gonna find this duct and this conduit right there, you know, and no surprises.

Robert: That that had to been your first shock to find out that that's not necessarily true.

David Basiji: That's that's exactly right. And and then and then the second second question was, you know, how do we how do we fix that? Why is it that way and how do we fix it? And you know, a lot of why it's that way is because fabrication requires, you know, that things sort of be different in the middle. So, you know, an easy example is you know, a field fit on a on a you you build a ski a skid, And it needs to go into the into the building under construction and and joined up with some some stubs that are in the wall or the floor or the ceiling, whatever. And you don't know exactly where it's gonna get placed. So you may want to field fit, you know, the the the output and inlet on that skid and say,

Robert: Yeah.

David Basiji: we add some extra length to this thing.

Robert: Yeah.

David Basiji: and we'll cut it in the field at the you know, we'll we'll drop it on the floor, we'll bolt it down, we'll know exactly where it is then, and then

Robert: Mm-hmm.

David Basiji: we'll we'll cut the pipe and weld to to suit. and so you add a foot, add six inches to these pipes. You don't want to put that in your BIM model. You know, that that's a fabrication detail. And in the end, that's not what it's going to look like. It's going to look like the BIM model. But in the middle of the process, it's got to look different. And so there's a lot of that that happens. And and so what we ended up doing in addition to getting all the on all these other tools with our tool control software is we built plugins for Revit and AutoCAD and Teclostructures and SDS2. Because we, you know, we want to cover the the the waterfront here with with whatever tools our customers are using. And those tools, they don't add any design features to Revit or whatever they're, you know, or AutoCAD, whatever one you're using. What they do is they make it easy to say what you want to fabricate, drawing a box around it or selecting it from an assembly list. And then it will analyze what you're pulling out of that model. And it'll it'll say, there's a little FF on this on this pipe end. That means field fit. So we're gonna add six inches to it. And when it so on its way out through our plugin and into our cloud, we make those modifications. And then when it gets to the to the pipe cutter, it does that. you know, it adds the length or or maybe it's you know your earlier point, you know, there's a sequence here. There's cutting and then there's bending and that sort of thing. You know, that bender is going to stretch the pipe when it bends it, right?

Robert: That is right. Mm-hmm.

David Basiji: ideally, your pipe cutter knows that. So it can it can trim a little bit. So after it goes through the bending process, it comes out to the right length. So there's a lot of, you know, a lot of benefits to having, you know, all these machines talking to the same system and to each other. so that they can coordinate through the process and and what comes out the other end, you know, is exactly what the field needs. And and then when, you know, when they're done with it, you know, it's exactly what the BIM model said it was supposed to be, or as close as you can, you know, reasonably get.

Robert: close as you can get. Yeah. Yeah, the the whole the whole world of exact is is again, I I would suppose in you know, in bioengineering, Exact is where you gotta where you gotta be. But when you were talking about, you know, concrete and steel and and things like that in these environments where it's a hundred and twenty degrees at one point and then the next, you know, and and three months later it's, you know, thirty degrees outside or whatever, like you've got to have areas where it's like, hey, we got we need a little wiggle room here. and we've got to be able to factor that in every time.

David Basiji: Right. Right.

Robert: I love it in the fact where you can get to a point where you can take like some of the historical information on based upon jobs of of similar nature and go, Hey, if we apply this here, this is kind of similar to what we did over here. We apply kind of some of these principles. We take this. It kind of gives you a little bit of early warning. Again on that field fit kind of thing that says, hey, by the way, the way we've done this best in the past is we've had to field fit it, right? We gave ourselves a little extra, we gave ourselves a a little wiggle room for this and this is how much we gave it, right?

David Basiji: Yep. Yep. And if you go to these MCAA meetings, or or the any of the the other, you know, meetings. Yep, yep, you'll you'll

Robert: Yeah. Maccus Macna and Nika

David Basiji: hear you you'll hear this story. They don't say it talk about it anymore 'cause they learned their lesson. But for a while there people got really you know sort of blinded by science there and and were like you know, 'cause you can you can cut your your all thread to the exact length that the model says it needs to be, accounting for floor slope and everything else. And and and so people tried that and found that

Robert: Yeah.

David Basiji: every every every hanger was totally different. How do you find it on the rack? How do you find it in the right place? Right? y and and you know sometimes exact is the wrong thing to do, right? And Especially, you know, if the if the system is set up to allow some some flexibility. You know, you can you can you can you don't need but two, three inches, you know, of accuracy on on the all thread. And if you do any less than that, you you're shooting yourself in the foot.

Robert: Yeah.

David Basiji: and so yeah, it it's it's one of those things where where you gotta you gotta find the right solution for the j just because you can doesn't mean you should.

Robert: I always say I I always say, you know, I say perfect doesn't exist exist, but excellence is part of the equation. Right. I mean we

David Basiji: Yep. That's a good one.

Robert: we do need excellence, when when when it comes to this kind of thing, but there's no such thing as perfect. Can you if you're trying to design for perfect, you're never going to be finished. so again, is that get the Parritos eighty twenty rule or whatever? That's that's the always the way that I look at it. Like you're Never gonna be perfect, but you know, excellence can be a part of the the equation in there, and that's what we want to really truly measure is is that kind of excellence portion, right? That's what

David Basiji: That's right. That's right.

Robert: we really want to measure. I've got I mean, age of AI is upon us, David.

David Basiji: Mm.

Robert: and I would imagine that Pipe Server is very, very busy. Especially when we start talking about needing prefabrication, needing these data centers and these behind the meter power systems to start really kind of turning heads and and actually start producing something. But I'd like to know, are you guys involved in any of that or is it is that kind of people you know, giving you phone calls going, Hey, I need to see what you guys got going on?

David Basiji: It is it is very busy. the are are you asking specifically about how we're how we're using AI or are you asking you know

Robert: That's no no, it's just it's just the the business of prefab and the business of of being able to, you know, utilize software's construction tech in order to improve processes because it seems like that's the biggest obstacle that's in front of us. It's not that we can't build these things, it's just that our processes suck and we don't and it and then we wonder why We wonder why we're, you know, eighteen months behind on this or, you know, on these different projects and we're we're wondering why we're doing that. It's because our processes are terrible.

David Basiji: Yeah, yeah. Yeah, yeah, we're you're I mean that's a very good point. The the the thing that we see a lot is that you know the process is there's several silos, right? There's the design and VDC, there's there's the fab shop, you know, there's the field, and there may be subcategories within there, but but typically things break down in the handoffs, right? From one from one to the next. Or if they don't break down, they're inefficient, right? and So you know, the way we've tried to deal with that is you know, we've developed a a cloud infrastructure. When when it first started out, it was really all about making it easier for the customers to get data from from CAD or BIM to the machine. that was very simple idea. Just we just instead of outputting a file and you know walking it over the machine on a USB stick or something because a lot of companies don't want to put their machines on their network 'cause it's a security risk until they realize a USB stick is actually a bigger security risk. But you know i

Robert: They did go through SOC two compliance and so they're like, no, I can't use this, I can't use that anymore.

David Basiji: Yep, exactly. And and you know, there was a when I first joined Pipe Server there was wave of ransomware and in all the in all a lot of our customers and it w it was a nightmare. But so so IT is really locked down now. And and and so it i that made these these issues even worse. 'cause you now, you know, how do you get stuff to the machine? and so when we first started out, you know, nobody wanted even an internet connection in the shop, right? It it was just what am I gonna do with that, right? and now it we we tell people you all you have to do is put your com give your compu your machine access to the internet and it can get to our cloud. It never has to even know about your corporate network. There's no security risk at all. and so so that was sort of the the address how we initially addressed the problem of getting data digitally from CAD or BIM to the machine is just set up a simple cloud, works over the internet, a couple basic plugins for for the design tools. define what you want, you know, make your your fabrication tweaks to it on its way out, and then get it to the machine. And then the machine will update the cloud in real time and say, this has been cut. These are almost ready, you know, so there's visibility there. And it does some nice things like protecting you from accidentally using a down rev version of a design. You know, it's always got the latest one. if you try and cut something twice, it'll it'll say, hey, you can't do that unless you, you know, ri hit click, I'm absolutely sure, three times, you know, all that stuff. So all the all the little you know gotcha yeah become easy to deal with once you've got the cloud connection to the machine.

Robert: Yeah.

David Basiji: and then once we had that, it was like, all right, well, let's start. working on, you know, more of of these issues. Let's let's not just focus on getting data from from the design to the machine. Let's build a whole work project tracking system, fabrication tracking system. Because there's a lot of stuff that, you know, people need to be able to do. They they need to know more than just, you know, is it is it is it ready or cut? You know, they need ever people have workflows. They it you know it's it there may be a 10 step process in their in their you know, in their fabrication journey, right? It goes from from design through, you know, detailing through approved for fabrication to in the shop to you know on the truck to received in the field to you know put up in the building to inspect it you know every shop has a different workflow and but the round you know the round shape is is is similar so we we set up we built on top of this foundation a a full project tracking system. So you can you can define based on all the statuses that you that you want. You you can say, okay, I want, you know, all the status list I just mentioned, or maybe there's a few extras in there. Maybe the maybe things go to paint, you know, in the fab shop. Or s some things go to paint and other things don't. What

Robert: Yeah, that's always the biggest thing on that. It's like, hey, why didn't this go to paint first or go to paint last? Right? The fish is, you know, we don't have a finish coat on it. Why is it sitting here on my job site and didn't have a finish coat on it? It's like I don't know.

David Basiji: Right, right. Exactly. And and and you know, weld locks, you know, if if you can digitize that, it's you know, that's huge. and you know, so so what we did is we we set up the system so you can you can define your workflow based on all these statuses. You can name them whatever you want, you can have as many as you want, you can set up rules for which statuses are allowed to flow to which next statuses. You know, maybe maybe some make sense and some don't, so you can block out the ones that don't. You can you can have reversion rules. So if if something went wrong and you need to back up, where where does it go back to? you can set up notifications.

Robert: Are you utilizing are you utilizing AI to kind of help notice these bottlenecks or or no?

David Basiji: No, we're not. I'll tell you how we're using AI in a minute, but but this is also deterministic that that you don't need or want AI. It's it's massive overkill. And the other thing is that you you you have to make sure that that there's you know AI is a is a statistical process, right? It's a it's a statistical tool. And if you've got a well defined process you know, those statistics may be very rock solid. It it may be give you the same answer nine hundred and ninety-nine times out of a thousand, but you can't guarantee it's going to give you the same answer every single time. So it's

Robert: That is right.

David Basiji: you know, it's great for large data sets where you are trying to pluck something unknown out of it and saying, tell me what is the most likely answer to this question based on all this big data you have access to.

Robert: Right. Right. What's the b yeah, what's the bell curve on this kind of thing? Like what's what's the most common thing I'm gonna find out of this whole

David Basiji: Exactly.

Robert: mess of information that I could never go through personally, but what's the most common thing I'm gonna find? What's the common thread, common theme? But I I t to your to your point, the precision that's needed with here ne needs to be well defined. I think this is a very, very interesting thing because I deal with this every day when I talk to people about their their productivity and their workflow. It's like y if you don't have your workflow defined, if it changes every day from and doesn't matter what you're doing, if your workflow continues to change because of some sort of exterior element that that is in your your control but yet you don't control it, that's your

David Basiji: Mm-hmm.

Robert: process problem in the way. Right. And I

David Basiji: That's right. That's right.

Robert: I that everybody wants to talk about how much productivity has increased in the manufacturing and you know, the fabrication space and how it's going to help construction is like, well, but the only thing that's gonna help construction, and I th that's the world that I concerned with, is them truly getting a grip on what their workflow really looks like. Right. You

David Basiji: Mm. Mm-hmm.

Robert: know, as simple as it gets, you can't just say that you're hanging drywall. There's so much more to it, right? Than just say, well, I'm gonna put some hang I'm gonna hang some drywall on the wall. Really? Okay. Which wall? How big? You know, how many sheets? You know, how many people is it gonna take? Where's it gonna, you know, where's it gonna come from? Where are you gonna stack it? All these kind of things. People don't do that. And then they wonder

David Basiji: All and

Robert: why they get behind.

David Basiji: and and you know, and if you're talking about you know, AI driven robotics for stuff like that, it's it there's a whole nother level of all right, what's the most efficient place to start? You know, what

Robert: Yeah.

David Basiji: what you know, what is that what does the nitty-gritty process look like, you know? I mean the basics, you know, hang it, you know, tape it, mud it, that that's easy. But you know, you get into a a room, maybe it's got some cove ceilings, maybe, you know, there's other weird stuff going on. You know, that that is where you know, human expertise is is you know, is extremely valuable. And that I think if it can be, you know, captured in data is an area where AI might might add some value, you know, is is in in a in a robotic kind of approach to that sort of thing. But for something as as you know, like you said, it's a it it's not a manufacturing process. in the sense that you're making the same thing over and over again, but it is a manufacturing process in that it's very well defined. You know all the steps, you know the order they have to be done in. They they're, you know, the the details of the parts you're making will vary, but the process won't. And so for that kind of thing, we don't we don't use AI. We just we s we say define it once. and and maybe maybe different jobs or different shops will will have, you know, variations on that. So you can copy it and modify it and that sort of thing. But fundamentally it it's deterministic. So we don't we don't use AI with it.

Robert: And so and and so I have another company that I run, Stentorian Research, where I I I make AI agents and I do a lot of research with these agents. And you know, the things that I'm finding out I find extremely interesting about this is that no matter how strictly you define your prompt, It ha it still has a tendency to wander, you know.

David Basiji: Mm-hmm.

Robert: A lot of times I feel like it's I call it A D D theater where it's like it's like where where is your mind going? I know it doesn't have a mind necessarily, but I'm just like, where are you going with this? And I you kinda gotta pull it back in. I have yet to find, you know, as in my work with it, I have yet to find anything that makes it feel any different than dealing with a new project engineer on a job.

David Basiji: Mm-hmm.

Robert: Right? Where you truly have to define exactly what you need and then you gotta check on it. It's it's that old adage, it's like and I don't care if it's AI or if it's a new work or whatever it is, but employees do what you inspect and not what you expect. It's the thing. But AI does the exact same. And I I don't think I should expect anything different because Current AI, current models are trained on human behavior and human knowledge. I

David Basiji: Mm-hmm.

Robert: I don't know, I'm I'm skeptical. You know, I I like to work with it. I think it's very, very interesting, and I I utilize the tools the best I can, but I'm I'm learning a lot about it, and I'm learning that, you know, you're gonna have to probably, you know, put in a lot of work to get the results that you want out of it. Which is

David Basiji: Yep. Yep.

Robert: kind of similar what to what you're talking about, is Even with pipe server. There's some work involved, g folks. You're gonna have to go through your workflow process and you're gonna have to define what comes next and what comes next and what comes next and be precise.

David Basiji: Yep. Yep. Exactly. And the you know what what what I find interesting is is, you know, to to your point about AI is you it it's both statistical in terms of what it's gonna give you, but it's also it's also bound by the data it has to work with.

Robert: Mm-hmm.

David Basiji: So and you know, if that if that data set was created by humans On average, it's going to give you the average human reaction or or phrase or whatever it is you're trying to you're asking it for.

Robert: Yeah.

David Basiji: you know, it's going to give you what is dominant in the data set. And what's dominant in most data sets is sort of the average response of a person, whatever that may be. And sometimes that's exactly what you want. Right. We we use AI internally, for example, for certain coding processes, right? Where where there's a task, you know, I'll a a good example, you know, we needed to add a feature to to to our cloud and our all all of our software that that would allow you to do single sign on, right? So instead of having a unique login and password for everything, right.

Robert: You need to s yeah, a single sign on security, yeah. I got

David Basiji: And and so you know there are services that offer this, right? Google has a single sign-on service, Microsoft has one, Facebook has one, Apple has one, every you know, everybody's got all these big all these big companies that have their fingers everywhere offer single sign-on services. And and they've been enough software engineers have implemented single sign-on with these large providers enough times that there's thousands and thousands of examples of it. in the data set. So

Robert: That's right.

David Basiji: when you go ask AI, show me some example code for, you know, Google single sign-on, boom. It it and it's gonna be right. Or or it's gonna be really close to what you need. And with just, you know, and a and a good programmer will will know exactly what they need to do to, you know, to to make it do exactly what they want.

Robert: That's right.

David Basiji: And so we were able to take a you know a feature request from a customer who, you know, who wanted this you know way back when. and turn it around in in in you know in in a matter of days. And whereas in the old days it might have been a month. So it AI definitely has its uses there. The other area where where we haven't gone yet, but we're gonna go, is you know, with these cloud systems, you know, with all these different stages in the workflow and QR coded labels on the parts that are getting scanned at every every point in the process as they're as

Robert: Mm-hmm.

David Basiji: they're you know moving through it. there is so much data there about, okay, this part, you know, spent this amount of time at this station being assembled. and then it, you know, it waited, you know, a day and a half to get on a truck to the site. And then it waited at the site for this long, and then it waited to get put up. And, you know, every Every one of the I mean and and that's just scratching the surface. There's granularity way below that

Robert: Yeah.

David Basiji: in in terms of what happened in the process. And all that is captured in the in our cloud system. So there's a there's an enormous data set that's unique to to each customer

Robert: Mm-hmm.

David Basiji: and that can also be looked at over all customers in an anonymous fashion. So that's that, you know, in our in our view of the world. That is where the value lies for AI for us is you know, giving our customers you know the ability to say, okay, I've done 10 jobs, you know, with your system in place. go in and and identify the bottlenecks in my process. What is taking longer than it should? And what defines should should is you know what the the global, you know, how long is my stuff waiting to get on the into the container and out to the site. You know, my stuff's taken a day and a half, but you know, the anonymous data from all the customers is saying, you know, most of them are shipping it within six hours. Maybe I got a problem there.

Robert: So I got a I got a good story for you. So in my research that I do, one of the things I have found is that my my focus has been lately on data centers and behind the behind the meter power. And what I have found in the e in the data that and it's publicly available data, but it it's in the in the data that I found is you know, you're pr it th si six times more likely to find errors in process and and delays and those types of things because of who you're using to do the work rather than what you're using. So it's people are you know, or companies or things like that are more likely to show losses regularly as as the company over material shortages, things like that. because basically the logic is it's easy to blame the materials And and they do. Everybody it's it's real easy to go, well, we're waiting on such and such. Not the fact that we could have done this, this and this and changed our you know, kinda to what you were just saying. We didn't use the data to go, where was my bottleneck here? You know, did I did the truck show up and and it laid the stuff down in the wrong spot and then it took us a week to find it and then that's why it's delayed or was it because, you know, it truly didn't get out of the factory from, you know, a certain point to certain point. Yeah. I just find

David Basiji: Right, right.

Robert: that I just find it fascinating. I and I've always said, even when we started this whole AI thing, AI will make us realize how human we really are.

David Basiji: Yeah, absolutely. Absolutely.

Robert: Be because we don't want you know there's these weird things about ourselves and I don't get it. I mean I do I do it too, but it's like we don't we have such a hard time admitting that w we're kind of the biggest part of our our problem when it comes to this stuff. And and AI will w with all of its benefits, it will recognize quickly, probably most likely, you know, where those where those faults lie and and I've always said that you know, productivity, one of my big things, but is you know, we need to develop not necessarily faster, but we need to be more productive with the time that we have, right?

David Basiji: Mm.

Robert: and in one of the things, no matter how many studies I do on this, as soon as you tell somebody that the productivity that they're actually earning is not what they thought. Like it it just it it sets them off as this whole thing, no, we're better than no no, we were doing better than that. That can't be right. It's it's this weird thing is you put data in front of people and humans in s in specific, you put data in front of us and we go, No, that can't be right. We immediately

David Basiji: Yeah, yeah.

Robert: think that we know better somehow 'cause our feelings got hurt. But man, you put the data in front of them and then they go, That can't be you know.

David Basiji: Yeah, yeah. Yeah, it's it's you know, and there there's that old saying, you know, you you can't fix what you can't measure. And and so having that data is is really the first step. And you know, and there's the other thing is y you you brought up, you know, individuals and and and that sort of thing. I you know, we we've seen also things like you know, with the weld logs, we we can tell how long, you know, how how many inches per minute people are welding and and some folks are faster, some folks are slower. And and a lot of people get a little, you know, nervous about that. You know, is you know is is Joe gonna look bad because you know he's not doing as many inches per per hour as as as as Bill is, right? And but when you've got enough data, you you know you can look at it in more dimensions and say, okay, but you know Joe's Joe's, you know, pass rate, you know, his X-ray pass rate is is you know, is the best in the shop. Right. So

Robert: Right. And Joe maybe Joe's measuring maybe Joe's welding half inch steel and somebody else's measuring quarter inch steel. I mean, there's total difference in time required. I mean, again, there's many different points of perspective that have to be included in this analysis.

David Basiji: That's right. That's right. And and I think that's an area where where AI can can you know also help digest the data well and say, you know, what what defines productivity? Is it you you need to be comparing apples to apples, but you also need to be looking at the the whole process and and what matters is, you know, ultimately, you know, did Joe may be slower, but you know, his stuff didn't need as much rework. It you know, we don't we have lower lower risk in the field because because we're you know we're not going to discover something after the fact. you know, things like that. And and and having you know very rich data and the ability to to look at more than just you know one simple metric that that might look bad on the surface but really isn't. I I think that's going to be that's going to be useful as

Robert: Yeah.

David Basiji: well.

Robert: So I think we kinda went down this alley, and we've kind of talked about the you know, sometimes the human bottleneck and the manufacturing backlog kind of thing. I'm interested in we kinda started down this road, but you have seen a Are you working more with companies? Are the the demand kind of going up for this f because of companies more working on data centers and and these infrastructure projects? Or is it kind of across the board it doesn't matter, all these different companies just need this?

David Basiji: It it's both. so there's there's no d I mean we we we have relationships with almost every pipe profiler manufacturer in the world now and and and their their business is going b gangbusters because you can't cut pipe fast enough to to build these data centers at the rate that they want to build so we're benefiting from that, right? The the a large fraction of these machines that get built are running our software and so we see that directly. But we also see you know, we we actually with our cloud plus solution, we target you know, the mid-market and the smaller shops. And of course we also target the large ones, but we you know we've gone out of our way to design a system that has you know a a suitable level of complexity, meaning it's not too difficult to use. but it still it still hits all the all the major you know features that that these customers need to to digit digitize their workflow. but it's simple and it's affordable. And and and that's that you know that's really what's important. you know, if you look at other solutions out there that are similar to what we do, they're they're almost universally targeted at you know the E and R top 400, you know, say call it the top ten percent of the market. you know, companies that can afford, you know, thirty to a hundred thousand dollars a year in in software fees and

Robert: Yeah.

David Basiji: can afford to hire a person or two or three to manage that tool internally full

Robert: Yeah.

David Basiji: time.

Robert: I know, but I'm gonna tell you and and this goes with anybody. And I'm not to not to poke anybody I'm not poking the Turner constructions of the world out in the eye or anything like that, but i seventy percent of the market is I think where you're at, David, it's that's where the meat of the market is really as far as these smaller shops, mid l mid level shops, they're doing a lot of the work, but yet they they they're like the redheaded stepchild of the industry when it comes to software. And I hate that because this is where we need the most improvement, for that kind of that digital transition of information to to actually happen. Because I would say the a lot of the bottleneck that we see, especially when it comes to, you know, any of this prefabrication that we're gonna have to be experiencing and it's gonna get bigger and bigger. The market's only gonna get bigger and bigger. These smaller shops have got to be able to step into that digital level of being able to go, I'm gonna pull this information out of BIM and get it to my machines, you know, like a like a pipe server would do, right.

David Basiji: That's right. That's right. Well, and and all the way out to the field. The whole the whole, you know, soup to nuts, right?

Robert: Cre yes. Yes. Gotta get the builders the information, yes.

David Basiji: that's right. And and so we've we've really designed our system for those folks. Basically, you know, different shops have different thresholds, but I think you know what we've seen is a half percent to one percent of turnover as an investment in productivity tools, software for the shop. is kind of a a a a reasonable benchmark. So if you're talking about a you know a a a company that does a million dollars a year, you know, that's that's what, so ten percent of that would be would be a hundred K.

Robert: Hundred thousand. Yeah.

David Basiji: one percent of that is is is ten K. You know that that ten K you can you can buy a lot of productivity for that. you know I mean for for our you know our stuff you know, our starter pack is is five seats of of Cloud Plus plus an automation software for to automate one station, it's five grand, you know. and and if you're not actively, you know, editing the workflow or changing things, you know, if you're just if you just want to view stuff, you can have as many seats as of that as you want for free. Right. So so even these very small shops can afford you know a tool like ours. And and that's You know, we're you you did a you did a an episode with I think it was constructible back in February.

Robert: Yeah, yeah.

David Basiji: And and I I think, you know, that that one really resonated with me because they're they're they're in a different, you know, different space. they're you know, single family home construction and and that kind of thing. But it the philosophy was very, you know, similar to ours, which is these smaller

Robert: Very similar, yes, very similar.

David Basiji: guys, they need they need these tools. And the and the big the big providers are not serving them.

Robert: Yes. And that is I always say the when I talk to c you know, contact founders and I think of that kind of thing, it's like it's like consider your market. and and consider trying not to hook the whale. Because

David Basiji: Mm-hmm.

Robert: one thing about hooking the whale is is that you may get on the hook and that's great, but you're gonna it's gonna take you a long time to get something implemented with them. and Do you have that kind of holding cost, first of all? You know, or would you rather maybe work down a level, get something implemented in six months or three months, right? And get something working. So then now you've got some feedback, you've got some income, you've got and then you've got something that you can really learn from because it's Again, having a running product is one thing, in the shop kind of thing, you know, in in the in your in your in your sandbox, but it's a whole nother thing when you get it out there in the field and then you get real people working with the tool and then start really testing what this thing will do and where your improvements need to come from. And you're gonna get a hell of a lot better feedback probably from those mid level and small level folks than you are gonna get from, you know, the turners of the world, which are gonna eat you alive. you know, they're just gonna eat you because they're gonna eat your time alive and and going 'cause they've been through this process and and I'm like I said, I'm not poking the big guys in the in the eye or anything. I don't wanna do that. I'm just saying, let's serve an underserved market. I think that's very, very important. Very important.

David Basiji: Yeah. Yeah. And we've got our, you know, we've got our our our software in in in every shop, large and small, right? 'Cause the the the big ones run, you know, many, many pipe cutters and and so we're we're in there. whether it's as tool control software or as cloud plus or as both. but yeah, we we've definitely value the you know the the closeness that we get with smaller shops. you know, we've we've had a few customers say, Hey, can we do a weekly call, you know, while we're you know, while we're getting up to speed here. you know, and and for you know, it starts out as a training thing and and then it turns into, you know, six months down the road, it yeah, everything's fine. We don't have any questions, but you know, there's this one thing that it doesn't do that we'd we'd we'd love to, you know, have it do and we is you know, we s we start hearing that from two or three of them and it's like, Now we know what we got to do next, you know, and and

Robert: Right.

David Basiji: it's it's incredible intelligence. and then when we roll it out, you know, a week or two later in beta form,

Robert: In fact.

David Basiji: they're like, wow, this is amazing. You guys, you know, you guys turned it around quick and and then you get that that that beautiful feedback and and before long, you know, you've done two or three turns on it and and it's it's ready to go for everybody, you know. And we love that. Absolutely love that.

Robert: Yeah. Yeah, the high tide r raises all boats and I think that that's i it Again, you're only gonna get high tide if you're getting a lot of good feedback, right? And you start getting and you utilize the feedback that you're getting from your client base and to to improve your product, right? That's

David Basiji: Mm-hmm.

Robert: that's where it is. That's where and again it's a relationship, right? It's not I sold you a product, it's not transactional. It's it's very much a relationship and it and it will continue to be that. Well, David, man, you know, forty five minutes runs quick around here, for the for the Texas drive time episode as I call it. But yeah, fantastic talking to you today. how can folks get a hold of pipe server? just real quick.

David Basiji: Well our our website is pipeserver.com. That's P Y P-E, S E R V E R dot com. if if you want to shoot us an email, we're we're info at at pipeserver dot com. if you wanna throw a bunch of money at us, go to sales at pipeserver dot com.

Robert: Yeah.

David Basiji: But we can start out with info and we can, you know, with go through the process. but yeah, that's that's probably the best way and and if people want to reach me directly, it's David.basici at pipeserver dot com. My last name is spelled B as and boy A S is and Sam I J I.

Robert: Very good. Well, thank you so much for being with me today. folks, that's it for the Contech Exec for right now. We'll be back next week. I've got the folks from Cavtera. I've got a three-part episode from those folks. You're gonna find out all you need to know about Palantir. All right.

David Basiji: Hm. Interesting.

Robert: Well, until then, everybody be safe, and we'll see you soon.

David Basiji: Thanks, Robert, appreciate it.